D-Wave Extends Its Chipmaking Deal With IonQ-Owned SkyWater
A familiar Minnesota manufacturing team remains behind D-Wave’s hardware. Device quality and repeatability explain why the factory relationship matters.
.jpg){kind=link}
D-Wave will continue developing and manufacturing quantum chips with SkyWater’s Minnesota team under a multi-year agreement announced October 13. The companies say the extension preserves process expertise built over more than a decade and includes intellectual-property protections.[1]
SkyWater now belongs to IonQ, which completed the acquisition on July 31 and said the foundry would keep serving other customers under the SkyWater name. D-Wave’s extension illustrates that continuing supplier role: a subsidiary of one quantum-computing company is making chips for another.[1][2]
An established relationship, with undisclosed new terms
This was already a multiyear relationship before today. An amendment effective March 31, 2025, filed with the U.S. Securities and Exchange Commission, referred to pricing for 2025, 2026 and 2027. Its dollar figures were redacted. Calling this the companies’ first multiyear deal would therefore be wrong.[3]
Neither the new contract’s length in years nor production volumes, pricing or new processor specifications were disclosed. The announcement does not identify exactly how the new agreement changes the earlier filed arrangement.[1][3]
Why a familiar fab can matter to a qubit
A quantum annealer turns a mathematical optimization problem into a physical energy landscape. D-Wave’s hardware uses superconducting flux with adjustable interactions. Programmed biases favor particular qubit states, while couplings express relationships between them; the annealing process changes the balance between the machine’s tunneling behavior and that problem landscape. The goal is to return low-energy states corresponding to good candidate solutions.[5][6]
That translation is imperfect. D-Wave’s technical documentation explains that noise and control errors can make the chip implement slightly different coefficients from those requested. Individual qubits also vary because their dimensions and circuit properties cannot be reproduced perfectly. A small physical difference can therefore become a difference in the optimization problem the machine actually explores.[4]
Calibration helps bridge that gap. Engineers measure the processor and its cryogenic environment, then build models connecting the requested problem to the controls applied to the hardware. QubitWire’s analysis: preserving manufacturing knowledge has engineering value because process development and calibration must deal with the same real devices. Continuity can support improvement, but it is not itself a performance score.[4]
The performance evidence predates this deal
D-Wave’s 2025 Advantage2 white paper provides a useful example of the connection between fabrication and computation. It compared Advantage2 with the preceding Advantage generation on three-dimensional spin-glass problems whose numerical precision could be varied. The company attributed better solution quality to a combination of increased energy scale, longer coherence and lower noise.[6]
That was a vendor benchmark between two quantum-annealer generations. It was not a comparison with the best classical application workflow, and its anneal-time comparisons are not interchangeable with end-to-end runtime. Those distinctions still matter when manufacturing news borrows the language of quantum performance.[6]
What would make the extension consequential?
A separate company-reported maintenance-scheduling pilot claimed a reduction in lot delays of up to 30%. The release calls it proof-of-technology work; a further study will add technician and resource constraints. It does not establish an independently verified production gain or quantum speedup.[1]
For anyone evaluating that application, the useful questions are practical: what schedule was used as the comparison, were both methods given the same constraints, how much total computing time was required, and did the improvement persist during ordinary factory operations? A result on a simplified model can be valuable without answering all of those questions.[1][4]
For the hardware relationship, the next evidence to watch is equally concrete: repeatable device quality, usable-device yield and systems delivered against stated specifications. SkyWater’s September launch of a dedicated quantum manufacturing offering makes those questions timely. Its stated aim was to bring process development, integration and packaging considerations into quantum-device development earlier.[7]